Identification of 'Carbon Hot-Spots' and Quantification of GHG

Feb 14, 2011 - ... Johan Kuylenstierna , SC Lenny Koh , Simon McQueen-Mason ... Thomas Wiedmann , Harry C. Wilting , Manfred Lenzen , Stephan Lutter ...
0 downloads 0 Views 833KB Size
Download PDF
ARTICLE pubs.acs.org/est

Identification of ‘Carbon Hot-Spots’ and Quantification of GHG Intensities in the Biodiesel Supply Chain Using Hybrid LCA and Structural Path Analysis Adolf A. Acquaye,†,* Thomas Wiedmann,†,‡ Kuishang Feng,§ Robert H. Crawford,# John Barrett,† Johan Kuylenstierna,† Aidan P. Duffy,|| S. C. Lenny Koh,r and Simon McQueen-MasonO †

Stockholm Environment Institute, University of York, Grimston House, York, U.K. Centre for Sustainability Accounting, Innovation Way, York Science Park, York, U.K. § Sustainability Research Institute, School of Earth and Environment, University of Leeds, Leeds, U.K. # Faculty of Architecture, Building and Planning, The University of Melbourne, Melbourne, Australia School of Civil and Building Services Engineering, Dublin Institute of Technology, Dublin, Ireland r Management School, University of Sheffield, Sheffield, U.K. O Department of Biology, University of York, York, U.K.

)

‡

bS Supporting Information ABSTRACT: It is expected that biodiesel production in the EU will remain the dominant contributor as part of a 10% minimum binding target for biofuel in transportation fuel by 2020 within the 20% renewable energy target in the overall EU energy mix. Life cycle assessments (LCA) of biodiesel to evaluate its environmental impacts have, however, remained questionable, mainly because of the adoption of a traditional process analysis approach resulting in system boundary truncation and because of issues regarding the impacts of land use change and N2O emissions from fertilizer application. In this study, a hybrid LCA methodology is used to evaluate the life cycle CO2 equivalent emissions of rape methyl ester (RME) biodiesel. The methodology uses input-output analysis to estimate upstream indirect emissions in order to complement traditional process LCA in a hybrid framework. It was estimated that traditional LCA accounted for 2.7 kg CO2-eq per kg of RME or 36.6% of total life cycle emissions of the RME supply chin. Further to the inclusion of upstream indirect impacts in the LCA system (which accounted for 23% of the total life cycle emissions), emissions due to direct land use change (6%) and indirect land use change (16.5%) and N2O emissions from fertilizer applications (17.9%) were also calculated. Structural path analysis is used to decompose upstream indirect emissions paths of the biodiesel supply chain in order to identify, quantify, and rank high carbon emissions paths or ‘hot-spots’ in the biodiesel supply chain. It was shown, for instance, that inputs from the ‘Other Chemical Products’ sector (identified as phosphoric acid, H3PO4) into the biodiesel production process represented the highest carbon emission path (or hot-spot) with 5.35% of total upstream indirect emissions of the RME biodiesel supply chain.

1. INTRODUCTION There has been a growing interest in the use of biofuels as a sustainable replacement for fossil fuels over recent years. This has led to many countries, including the UK and the wider EU community, formulating policies that set out long-term strategies to promote biofuel production and use driven mainly by policy goals such as: reducing greenhouse gas emissions through the decarbonization of transport fuels, diversifying fuel supply sources, and developing long-term replacements for fossil oil. The EU has a long-term vision for biofuels, proposing that by 2030 and beyond, clean and CO2-efficient biofuels would make up 25% of the EU’s transport fuel needs.1 Refer to Figure S1 in the Supporting Information (SI) for an illustration of the transition plan of past EU policies affecting biofuels and the time scale for future commitments. Biodiesel is Europe’s dominant renewable fuel2 with rapeseed accounting for about 80% of primary feedstock for biodiesel processing and about 75% share of total oilseed production of r 2011 American Chemical Society

EU-27 in 2009-10.3 Production of biodiesel on an industrial scale began in 1992 about five years before the EU’s Energy White Paper ’Energy for the Future’, driven mainly by positive signals in terms of support from member states and the EU Commission. Figure S2 in the SI shows the trend in the growth of biodiesel in the EU since 1992. Despite the potential growth and benefits of biofuels, many research findings have raised arguments against them in a global context. The Food and Agricultural Organisation4 acknowledges that, although biofuels under certain conditions help reduce greenhouse gas emissions (GHG), the global effects of an expansion of biofuel production will depend crucially on where and how feedstocks are produced. It is therefore anticipated that Received: October 8, 2010 Accepted: January 13, 2011 Revised: January 9, 2011 Published: February 14, 2011 2471

dx.doi.org/10.1021/es103410q | Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology the more sustainable second-generation biofuels produced from nonfood crops and residues can provide the best opportunity for the commercial viability and development of the sector from 2014 onward.5 Lifecycle assessment (LCA) of second-generation bioethanol produced from surplus forest-bioenergy resources in Norway for example was estimated to potentially save 6-8% of Norway’s global warming GHG emissions associated with road transportation.6 The accelerating use of biomass including, cereals such as wheat, maize, sugar, and oilseed for biofuel and power generation has come about because of positive government directives and political decisions.7 However, the exact impact on the resource base and the environment due to the demand for biofuels is unknown. Many authors have therefore undertaken studies to evaluate the environmental impacts of biofuel production.8-10 These studies have mainly used traditional LCA methods based on ISO 14040 and mostly involved comparative studies with traditional fossil fuel production.11 Traditional LCA of biofuel production involves setting a system boundary for the biofuel supply chain and using process analysis data to estimate the carbon impacts of selected supply chains within the system boundary. It is, however, well recognized that because of difficulties in collecting process-specific data in LCA and the infinite number of possible supply chain paths, the use of hybrid LCA provides a more comprehensive framework for the evaluation of environmental impacts of upstream production.12-15 A hybrid LCA combines the specificity of process analysis with the extended system boundary of input-output (IO) analysis. Hybrid LCA has had many applications. Lenzen and Wachsmann16 and Crawford 17 demonstrated the use of a hybrid LCA technique in the assessment of the energy content of wind turbines in order to achieve system completeness. A limited number of studies using hybrid LCA have been undertaken on biofuels. Bright et al.18 undertook an environmental assessment of wood-based biofuel to estimate the cumulative global warming mitigation under different scenarios in Norway. The hybrid LCA in this study consisted of a two-region (Norway and the European Union) IO model and process analysis inventory for the biofuel options. In this paper the life cycle GHG emissions of a typical biodiesel supply chain are calculated using hybrid LCA, incorporating process-specific data of rape methyl ester (RME) production and inputs from higher upstream processes such as chemical inputs, mining, transportation, banking, equipment, etc., based on inputoutput analysis. Direct and indirect emissions in the biofuel supply chain are determined, including direct and indirect land use change and N2O emissions from fertilizer application. Furthermore, structural path analysis (a decomposition technique used in economic and ecological systems analysis) is applied to identify, quantify, and rank high carbon emission paths - or ‘carbon hot-spots’ - in the supply chain. Some studies have demonstrated the use of SPA as a decomposition technique of environmental impacts in a LCA context.19,20 To the best of the authors’ knowledge, this is the first time that SPA has been used for an analysis of the biodiesel supply chain. This detailed analysis is aimed at helping to tailor and prioritize mitigation efforts through the use of biofuels.

2. MATERIAL AND METHODS “Integrated hybrid LCA” as defined by Suh and Huppes21 is applied in this study. This form of hybrid LCA combines a process matrix and an IO matrix in a consistent mathematical

ARTICLE

framework.22 Whereas the process component systematically computes physical inputs and outputs of each production step within the system boundary, the input-output component completes the analysis by enumerating upstream indirect inputs from outside the process system boundary. For an integrated hybrid assessment of biofuel supply chains, the process matrix is linked to the input-output matrix using the operational expenditure of biofuel production to account for upstream inputs. As shown by Suh and Huppes,21 the general relationship for the integrated hybrid model is given in matrix notation by #-1 " # " #" Ep Ap 0 -D y ~ ð1Þ Phybrid ¼ 0 Ei-o -U ðI-Ai-o Þ 0 where P~hybrid = total (direct and indirect) environmental impact (e.g., CO2-eq emissions) associated with one unit of final demand y for the product (here biodiesel), Ap = square matrix representation of process inventory, (dimension:s  s), Ai-o = IO technology coefficient matrix (dimension:m  m), I = identity matrix (dimension:m  m), U = matrix representation of upstream cut-offs to the process system (dimension:m  s), D = matrix of downstream cut-offs to the process system (dimension:s  m), Ep = process inventory environmental extension matrix. CO2-eq emissions are diagonalized (dimension:m  s), Ei-o = IO environmental extension matrix. CO"2-eq # emissions are diagonalized (dimension:m  s) y = functional unit column matrix with dimension (s þ 0 m,1) where all entries are 0 except y. Matrix Ap describes the product inputs into processes as captured in the unit process exchanges (or process analysis inventory from ecoinvent in this case) and described in Table S1. These processes, together with the sectoral inputs from IO sectors, are used to draw up the biodiesel supply chain map as depicted in Figure S3. Matrix U, which is assigned a negative sign, represents the higher upstream inputs from the IO system to the process system. Matrix D, also assigned a negative sign, represents the (downstream) use of goods/process inputs from the process to the background economy (IO system). As explained by Suh and Huppes,21 the downstream cutoff matrix represents the link from the process-based (foreground) system to the IO-based (background) system. It can be argued that the downstream cutoff flows in D are often small compared to the - normally much larger - background economy (cf. ref 23). The aim of the present paper is to quantify the total emissions of biodiesel production in the status-quo economy of the UK in 2004 when the market share of biodiesel as a percentage of road transport fuel was at a modest 0.09% in 2004.24 For the sake of simplification we therefore neglect interactions with the background economy and set values in D are set to zero. We acknowledge, however, that a more general use of biodiesel in the economy would ideally be evaluated by including industries’ expenditures on biodiesel in D (e.g., by assuming different market penetrations of biodiesel in a number of scenarios). The final demand ~y for biodiesel also represents the functional unit of the LCA system, set to 1 kg of RME biodiesel in this study. In order to achieve a complete LCA system for the biodiesel supply chain, upstream cut-offs from the process-based LCA system were estimated using input-output analysis. For example, 2472

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology

ARTICLE

to estimate the contributions of an upstream service (for example: administration) for a given process inventory (for example: electricity) already captured in the process matrix, Ap the following steps were taken. The unit cost of the process under consideration (example: electricity) was obtained [£/kWh]. This was multiplied by the input (in physical terms) of electricity [kWh] obtained from the process matrix. The results, k (that is [£/kWh]* [kWh]) represents the amount of electricity (in £) needed to produce 1 kg of final demand of biodiesel. This amount is then used as a scalar multiplier to the column aij of the IO technology matrix, Ai-o where j corresponds to the Electricity industry. To avoid double counting, all inputs already captured in the process matrix are discounted from the resulting column vector kaij. The corrected values kaij * become elements of the upstream input matrix U. The administrative expenditure linking the process LCA electricity to the IO table corresponds to kaij where l corresponds to Administration as a product and j Electricity as an industry. Refer to Spreadsheet S1. Uncertainty in upstream emissions was estimated by including the maximum/minimum IO upstream cut-offs into the LCA system. To account for the maximum IO upstream cut-offs, all potential sectoral products that are indirect input requirements into biodiesel production are included. Similarly, to account for minimum IO upstream cut-offs, only sectoral products that are highly probable indirect input requirements into biodiesel production supply chain are included. Refer to the supplementary Spreadsheet S1 for inputs into the upstream supply chain for the maximum and minimum case scenarios. Besides its mathematical consistency, integrated hybrid LCA provides a comprehensive framework because all inputs associated with the biodiesel supply chain can be expressed by the combination of process and IO matrices. 2.1. Structural Path Analysis. Taylor’s series expansion is applied only to the IO part of eq 1 because the inputs of the unit process obtained from ecoinvent are clearly known PIO ¼ ½E  ðI - AÞ-1 y 0

¼ EIy þ EA1 y þ EA2 y þ EA3 y þ 3 3 3 þ EAn y

ð2Þ

EIy’ represents the direct GHG emissions emitted (at production level 0) for a given demand, and EAny is the indirect GHG emissions emitted for a given final demand at the nth production level. The Taylor series expansion of the Leontief inverse matrix can be further decomposed by unravelling the A matrix (or the IO technology coefficient matrix with elements, aij) into a series of structural paths at the nth order to systematically identify important supply chains.25 Summing up across all products i and industries, j, the total environmental impact of a final demand bundle yt can be decomposed into 0

PIO ¼

n X n X

ej ðIjt þ ajk þ

t ¼1 j¼1

þ

n X n X t ¼1 k¼1

n X

ajk akt

k¼1

ajt ajk akt þ 3 3 3 Þy1

ð3Þ

where ej is the emission intensity of industry j, and elements amm represent transaction coefficients between sector n and m. Each multiplied term represents the contribution of an individual supply chain path. In the case of biodiesel in this study, the emission ‘carbon hot spots’ in the supply chain were to be

identified and therefore the combined upstream inputs from matrix U as the demand bundle yi were used yi ¼

n X

kaij /

ð4Þ

j¼1

where k represents the £ equivalent needed by industry j to produce 1 kg of biodiesel. The decomposition of the series expansion can be represented as a tree diagram (refer to Figure S4) whereby each tier in the tree represents a different production layer and each node gives the contribution to total environmental impacts from the demand, y.26 Production layer refers to the stage of supply to the main product. Production layer 0 therefore refers to the biodiesel production process. Production layer 1 is the first stage of the upstream supply chain, and production layer 2 is the supplier to the first upstream supplier of the biodiesel process. In a SPA disaggregation of a product system, each node represents a contribution to the total environmental impacts from the demand, y. The maximum number of nodes at a production level is given by Number of Nodes ¼ nl þ 1

ð5Þ

where n = number of sectors in the economy, and l = production level. The importance of supply chain decomposition in disentangling upstream emission paths in a product system is evident in the fact that upstream environmental impacts are often greater than direct environmental impacts in a supply chain. In a carbon footprint case study of economic sectors in Australia and the US, Huang et al.27 for example, showed that direct emissions of the majority of sectors are below 20% of the total carbon footprint and can be as low as 1%. To maximize the potential for biodiesel to achieve real CO2 emissions reductions compared to fossil fuels, high emissions intensity paths or ‘hot-spots’ in the supply chain must be identified, and possible lower emission alternative processes for the production of biofuels must be found. 2.2. Process Analysis Data. The 2010 ecoinvent database v2.2 was used to compile the process analysis life cycle inventory described as unit process exchanges. This data set includes production of biodiesel rape methyl ester (RME) from rape oil from esterification plants in the EU. The operation of storage tanks and fuel stations, including the distribution to the final consumer and all necessary transport requirements are included. Emissions arising from evaporation and treatment of effluents (may also refer to the air emissions of the plant) are also included. For the analysis, corresponding CO2-eq emissions data of unit process exchanges, Ep, emitted in producing 1 kg of RME biodiesel were determined. Biogenic CO2 was captured in the unit process exchanges obtained from ecoinvent.28 It was calculated using the principle of carbon balance (input of carbon = output of carbon); that is, the uptake of carbon during plant growth plus all inputs of biogenic carbon with all preproducts minus biogenic carbon emissions should equal the biogenic carbon content of the biofuel or the product after all allocations have been done.28 The unit process exchanges representing the process analysis data from ecoinvent are presented as Table S1 as part of the Supporting Information. 2473

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology 2.3. Input-Output (IO) Analysis Data. Previously constructed 2004 UK domestic and UK imports supply and use tables disaggregated to 178 sectors were used to derive the input-output data used in the study.29 Wiedmann et al.29 describe the construction of a multiregional input-output (MRIO) model using UK national IO tables and rest-of-world (ROW) tables from the Global Trade Analysis Project (GTAP). A technology coefficients matrix was derived for both the UK domestic and UK imports use table. For the purpose of the present study, the ROW economy is represented as one symmetric table (technical details of this 2-region model have been described in ref 30). Columns of UK and ROW industry input requirements are augmented with data for greenhouse gas emissions to derive sectoral emissions intensities (kg CO2-eq/£) for the environmental matrix, Ei-o. A supply chain map illustrating the comprehensive system boundary framework of the biodiesel supply chain adopted in this study is available in Figure S3 in the SI. 2.4. Allocation Factors. The production of RME results in multiple product outputs. For example, the processing of oil mill into rape oil also results in the production of rape mill as a byproduct. The esterification of vegetable oil into RME also produces glycerine and potassium sulfate. In order to deal with multiple product outputs, LCA studies apply the method of either allocation or system expansion. In the first case, inventory data are allocated to the main product, byproduct, and waste, respectively, in order to assign material inputs and environmental impact. In system expansion, the boundary is extended to account for the input and output flows of all products. In this study, we use the first option, allocation, as we are specifically interested in the provision of biodiesel. Allocation factors can be based either on mass flow, energy value, or economic revenue of coproducts. Economic allocation has been established as a recognized way of systematically executing allocation in LCA.31-33 The International Standards Organisation34 also gives this allocation option in Step 3 of its allocation procedure. Hence, in this study, the economic revenue allocation as adopted in ecoinvent28 was used. To reduce the uncertainty related to economic allocation because of potential fluctuations in the economic values of product and coproducts, the environmental burdens are allocated according to the revenue of all process products, based usually on the average prices for three consecutive years (refer to Table S2 for allocation details). Allocation factors for other methods related to the production of RME are also presented in ecoinvent.28

3. RESULTS AND DISCUSSIONS 3.1. Hybrid Life Cycle CO2-eq of Biodiesel Production. The total emissions of all unit process exchanges representing the process analysis data of the biodiesel production process is 2.7 kg CO2-eq or 36.6% of the total life cycle emissions. IO upstream indirect emissions (for the base case impact scenario) account for 1.7 kg CO2-eq of the total life cycle emissions. Upstream emissions include embodied emissions such as those associated with utilities, equipments, chemicals, mining, construction of buildings, maintenance, services such as banking and finance, insurance, research, and development, advertising, etc., and accounted for approximately 23% of total emissions. A further breakdown of these emissions is provided in Section 3.3 (Structural Path Analysis of Biodiesel Supply Chain Emissions). Refer also to Table S3 and Figure S6.

ARTICLE

It was also estimated (from the process analysis inventory in ecoinvent) that the esterification of vegetable oil to RME process accounted for 35.5% of the total emissions or 97% of emissions due to the unit process exchanges in the process inventory. The other unit process exchanges: road and train transport, electricity supply, regional distribution of oil, waste management and water treatment from the process analysis inventory in ecoinvent collectively accounted for around 1.1% of the total emissions associated with the RME biodiesel production process. 3.2. Other Impacts. It has generally been argued that greenhouse gas releases from land use change and nitrous oxide (N2O) emissions from the use of fertilizers can potentially be significant enough to change the environmental profile of biodiesel.35,36 With N2O having a global warming potential 298 times that of CO2 when considered over a 100-year period,37 the use of nitrogen fertilizers has the potential to significantly affect the GHG emissions balance of biodiesel. N2O is emitted both directly from soils due to the use of nitrogen-based fertilizers and microbial transformations of organic nitrogen (N) and also indirectly with nitrogen losses through volatilization, leaching, and runoff of N-compounds that are converted into N2O off site. Also, the European Commission Joint Research Centre,38 Searchinger et al.,39 and Fargione et al.40 have all stated that indirect land-use change could potentially release enough greenhouse gases to negate the savings from conventional biofuels. Land use can be defined as the type of activity being carried out on a unit of land and the change in land use can be either direct or indirect. Direct land-use change occurs when feedstock being cultivated for biofuels production (e.g., rapeseed for biodiesel) displaces a prior land use (e.g., forest), thereby generating possible changes in the carbon stock of that land. Indirect land-use change on the other hand occurs when pressure on agriculture due to the displacement of previous activity or use of the biomass induces land-use changes on other land.11 3.2.1. Estimation of N2O Emissions from Fertilizer Application. The Intergovernmental Panel on Climate Change, IPCC41 estimate that the direct emission factor associated with N2O emissions is likely to be 1% of the N applied to the soil. Jungbluth et al.28 also estimated a direct emissions factor of 1.25% of the N-input and an indirect emission factor of 2.5% from the nitrogen that is leached as nitrate. Crutzen et al.42 however state that global analysis of N2O emissions had been previously underestimated and shows that N2O emission factor (direct and indirect) in agro-biofuel production is 3-5% of N applied (that is 0.03-0.05 kg N2O-N (kg N)-1. The maximum value of 5% in Crutzen et al.42 is used to assess the highest impact case scenario in this study. The minimum value of 3% Crutzen et al.42 is applied to determine the uncertainty range between the maximum and the minimum impact scenario. Refer to Smeets et al.43 for additional information on the contribution of N2O to the greenhouse gas balance of first-generation biofuels. The fertilizer input rate assumed in this study (137.4 kg-N ha-1 yr-1) was used in the JEC - Joint Research Centre-EUCAR-CONCAWE collaboration on biofuel program.44 In the cultivation of 1 kg of rape, the following conditions were assumed: i Land Use: Transformation from nonirrigated arable land accounted for 71% of land use at 2.08 m2 per kg of rape while transformation from pasture and meadow land accounted for 29% of land use at 0.85 m2 per kg of rape45 ii Land Occupation: 11 months per year permanent land-use occupation.46 2474

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology

ARTICLE

Figure 1. Life cycle emissions of RME production and supply chain.

iii It was deduced from ecoinvent28 that 2.6 kg of rapeseed is required to produce 1 kg of RME biodiesel The relationship used to determine the life cycle CO2-eq emissions associated with the use of nitrogen fertilizers can be expressed as Life Cycle Emissions due to N2 O Use ¼ ef yfi rs kn ðGWPÞN2 O

j X

li pi

ð6Þ

i¼1

where ef = N2O emissions factor [kg.N2O-N per kg-N], y = time between planting and harvest of the bioenergy crop [years], fi = fertilizer input rate [kg-N ha-1 yr-1], rs = ratio of the kg of rapeseed to required to produce 1 kg of RME, kn = factor to convert from N2O-N to N2O [equivalent to (44/28)], (GWP)N2O = Global Warming Potential of N2O, l = area of land occupied by bioenergy (biodiesel) crop [m2 per kg of rapeseed], and p = ratio of the area of particular land type occupied by bioenergy crop to the total area of different land used (in cases where only one land type is used, p will be 1). Using a GWP of 298 for N2O and effective land size of 1.723 m2 or (1.723  10-4 ha) per kg of rape seed, it is estimated from eq 6 that the N2O emissions contributed between a high impact scenario of 1.32 kg CO2-eq per kg of RME and a low impact scenario of 0.80 kg CO2-eq per kg of RME biodiesel. 3.2.2. Direct and Indirect Land Use Change. The Intergovernmental Panel on Climate Change47 reports that due to direct land use change, changes in carbon stock per hectare of bioenergy crop cultivation occurs in the following carbon pools for cropland: biomass (above ground biomass and below ground biomass); dead organic matter (dead wood and litter); and soils (soil organic matter). The total change in carbon stock is calculated using eq 7 below10   tC ΔCarbon Stock Soil ha   X tC ¼ Carbon Stock Change Factor  Ti ½yr ð7Þ hayr i where Ti = time between planting and harvest of the bioenergy crop [years]. Based on data on carbon stock change factors for the carbon pools of cropland from the IPCC Guidelines for National Greenhouse Gas Inventories - Agriculture Forestry and other

Land Use,47 direct land use change was estimated to be 0.44 kg CO2-eq per kg of RME. Indirect land use change (iLUC) is calculated using the theoretical global average indirect land use change factor.48 In this study it was assumed that the cultivation of rape seed occurred on 71% arable and 29% pasture and meadow land. A ‘maximum risk’ or ‘maximum iLUC order of magnitude’ representing a 75% share of nonzero risk biofuel is assigned an iLUC factor of 15 t CO2-eq/ha/yr, while a ‘low risk’ or ‘low iLUC order of magnitude’ representing 25% of all nonzero risk biofuels are subject to theoretical full iLUC factor of 5 t CO2-eq/ha/yr.48 The term risk refers to the level of impact due to the conversion of food crop land into bioenergy crop land. A low risk biofuel is therefore assumed to be produced from feedstock cultivated on set-aside or abandoned land. By weighting these iLUC factors according to the ratio of land type and land sizes assumed in the cultivation of rape seed in this study, the iLUC factor used for to the production of 1 kg of RME biodiesel and its coproducts is estimated to be 11.8 t CO2-eq/ha/yr or 1.86 kg CO2-eq. Taking into account the allocation factors in Supporting Information Table S2, iLUC is calculated to be 1.22 kg CO2-eq per kg of RME. Uncertainty in the impact of land-use change refers to the variability of indirect land-use change factors due to the type of land used in the cultivation of feedstock. Based on the assumptions for maximum/minimum iLUC risk referred to above, the uncertainty range for iLUC for producing 1 kg of RME biodiesel is estimated to be between of 0.52 to 1.55 kg CO2-eq. The emissions associated with all stages of the RME biodiesel production are shown in Figure 1. The total life cycle CO2-eq emissions for 1 kg of RME biodiesel were calculated to be 7.38 kg CO2-eq or 199 g CO2-eq/MJ. By accounting for uncertainty in the assessment, it was estimated that the results are in the range 5.03 to 8.44 kg CO2-eq per kg of RME biodiesel. Refer to Figure S5 for normalized results in energy units. 3.3. Structural Path Analysis and Hotspots of Biodiesel Supply Chain Emissions. In Figure S6, the cumulative impacts of sectoral emissions from the higher upstream supply chain paths of biofuel production are presented giving an indication of the relative contribution of each IO sector. These higher upstream supply chain paths represent the IO component of upstream inputs. Seven production layers of RME biodiesel were analyzed. It was estimated that the ‘Utilities Sector’ was the highest sectoral emitter accounting for 172 g CO2-eq or 44.5% of total upstream emissions. This was followed by the ‘Chemical 2475

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology Sector’ emitting 90 g CO2-eq or 23.3% of total upstream emissions. As can be seen from Figure S6, the next four sectoral emissions were ‘Transportation and Communication’ (37 g CO2-eq or 9.6%), ‘Mining’ (21 g CO2-eq or 5.4%), ‘Minerals’ (19 g CO2-eq or 4.8%), and ‘Fuels’ (14 g CO2-eq or 3.7%). Structural path analysis (SPA) is used to show the interconnections of various products and industries within the biodiesel supply chain and identify, rank, and estimate the CO2-eq of the high emissions intensity paths or ‘carbon hot-spots’. 150 of the most important paths of the biodiesel supply chain were extracted in the SPA. The cutoff threshold for individual path contributions was set at 0.05% of total impacts in the analysis of the supply chain paths. Detailed results for each of the top 50 paths are shown in Table S3 in the SI. It was found that CO2-eq emissions impacts on the biodiesel supply chain originate across the entire economy but of the top 150 paths, the majority originate from the sectors ‘Other Chemical Products’, ‘Organic Basic Chemicals’, ‘ElectricityCoal’, ‘Distribution and Trade’, ‘Electricity-Gas’, and ‘Freight transport by Road’. The “hottest spot” or the highest carbon intensity path of the biodiesel upstream supply chain was identified as a path order 1: Rest of World (ROW) Sector (102) ‘Other Chemical Products’ > Biofuel Process with an estimated 20.7 g CO2-eq or 5.35% of the total emissions. This path describes the emissions chain: ‘Other Chemical Products’ used as an input in the biodiesel production process.

4. DISCUSSION The life cycle assessment of the RME biodiesel supply chain estimated the total life cycle emissions of biodiesel production to be 7.38 kg CO2-eq per kg with an uncertainty range of 5.03 to 8.44 kg CO2-eq per kg. The uncertainty was a result of variability in indirect land use change, N2O emissions and IO higher upstream emissions. IO higher upstream emissions accounted for approximately 23% of total CO2-eq emissions. The use of a hybrid method ensured the integration of process and IO analysis such that higher upstream inputs into sectors such as utilities, transportation, chemicals, mining, services, etc. which are normally excluded from traditional life cycle assessments, are taken into account. In contrast, Halleux et al.,8 Hoefnagels et al.,10 and Kim et al.49 all undertook life cycle assessments of biofuels using traditional LCA but did not account for upstream emissions outside the process system boundary resulting in the truncation of the product system. Given that some past assessments of biofuels have also neglected the impacts of land use change and N2O emissions, assuming that the impacts of land use change, N2O emissions and IO upstream emissions cut-offs were truncated from the biofuel product system, as has previously been the case, the total emissions would have been 2.7 kg CO2-eq per kg. This would have resulted in a 63% underestimation of the total life cycle emissions. Therefore, IO upstream emissions cutoff, N2O emissions, and land-use change represent significant impacts which are determinants that can change the environmental profile of the biodiesel supply chain. It was observed that the lack of process-specific data increases uncertainties in life cycle assessments. The uncertainty estimates for this study were based on data variability in indirect land use change, N2O emissions from fertilizer application and aggregation of IO data, resulting in the estimation of minimum and maximum carbon impacts of the RME biodiesel supply chain. The estimation of emissions was

ARTICLE

based on economic allocation between the RME biodiesel supply chain and coproducts. This has been recognized as one way of systematically executing allocation in LCA.31-33 Structural path analysis (SPA) is useful in describing and characterizing carbon hotspots in the supply chain. Specific processes in the RME biodiesel supply chain can be matched to the structural paths in order to identify the hotspots in the supply chain. For example, the first ranked structural path: Other chemical products > Biofuel Process can be identified as the inputs of industrial grade phosphoric acid, H3PO4 (85% in water) into the biodiesel production process. Likewise, the second ranked structural path: Organic basic chemicals > Biofuel Process describes the inputs of methanol into the biodiesel production process. As can be seen from the ranked structural paths available in the Table S3, all the paths end as a direct input into the RME diesel production process. The demanding sector is therefore responsible for the emissions caused, but the emissions might occur upstream of that sector. For example, in the 12th ranked structural path: UK- Electricity - Coal > UKDistribution and Trade in Electricity > Biofuel Process; the biodiesel production process is responsible for emitting 3.28 g CO2-eq per kg of RME biodiesel although it occurs upstream of the production process. SPA provides a unique way of identifying processes in the entire supply chain with hot-spots thereby ensuring that appropriate intervention measures and effective policies can be prioritized and implemented to reduce carbon impacts. Emissions resulting from industries in the ROW indicate that biofuel energy policies should not be limited to the UK but rather a holistic approach should be adopted to account for emissions occurring beyond the boundaries of the UK. SPA for coproducts was not undertaken since system expansion allocation was not used As has been demonstrated for RME biodiesel, a systematic analysis of hybrid LCA and application of SPA should also be extended to second generation biofuel because the environmental impact of second-generation biofuel production can vary considerably depending on the conversion route as well as the feedstock and site-specific conditions (refs 50 and 51). This is because the benefits of second generation biofuels is being promoted (e.g., refs 1 and 52) but the environmental profile is not fully understood.

’ ASSOCIATED CONTENT

bS

Supporting Information. Further tables (unit process exchanges, allocation, and SPA results) and figures (biodiesel in the EU, biodiesel hybridized system boundary, results) and spreadsheet of input-output analysis. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: þ441904322893. Fax: þ441904322898. E-mail: adolf. [email protected].

’ ACKNOWLEDGMENT We gratefully acknowledge the financial support provided by the Centre for Low Carbon Future (CLCF), York, UK. The structural path analysis was performed by using the Triple 2476

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology Bottom Line tool developed by the Centre for Integrated Sustainability Analysis (ISA) at the University of Sydney and supplied in the UK by the Centre for Sustainability Accounting (CenSA), York, UK (http://www.bottomline3.co.uk).

’ REFERENCES (1) European Commission, E., An EU Strategy for Biofuels, COM(2006). 2006, 34 final. (2) van Thuijl, E.; Roos, C.; Beurskens, L. An overview of Biofuel Technologies, Markets And Policies In Europe; 2003. (3) Agricultural Commodity Prices EU-27 Oilseeds Production http://www.agricommodityprices.com/futures_prices.php?id=154 (accessed 15th June). (4) Food and Agriculture Organization, F. A. O., The State of Food and Agriculture 2008. 2008. (5) European Commission, The impact of a minimum 10% obligation for biofuel use in the EU-27 in 2020 on agricultural markets. In 2007. (6) Bright, R. M.; Strømman, A. H. Life Cycle Assessment of Second Generation Bioethanols Produced From Scandinavian Boreal Forest Resources. J. Ind. Ecol. 2009, 13 (4), 514–531. (7) Spiertz, J. H. J.; Ewert, F. Crop production and resource use to meet the growing demand for food, feed and fuel: opportunities and constraints. NJAS - Wageningen J. Life Sci. 2009, 56 (4), 281–300. (8) Halleux, H.; Lassaux, S.; Renzoni, R.; Germain, A., Comparative life cycle assessment of two biofuels ethanol from sugar beet and rapeseed methyl ester. Int. J. Life Cycle Assess. 2008, 13, (3). (9) Ou, X.; Zhang, X.; Chang, S.; Guo, Q., Energy consumption and GHG emissions of six biofuel pathways by LCA in (the) People’s Republic of China. Appl. Energy 2009, 86, (1). (10) Hoefnagels, R.; Smeets, E.; Faaij, A. Greenhouse gas footprints of different biofuel production systems. Renewable Sustainable Energy Rev. 2010, 14 (7), 1661–1694. (11) Gnansounou, E.; Panichelli, L.; Dauriat, A.; Villegas, J. D., Accounting for indirect land-use changes in GHG balances of biofuels: Review of current approaches. Working Paper 2008, REF. 437.101, EPFL. (12) Lenzen, M. Differential Convergence of Life-Cycle Inventories toward Upstream Production Layers. J. Ind. Eco. 2002, 6 (3-4), 137– 160. (13) Crawford, R. H. Validation of the use of input-output data for embodied energy analysis of the Australian construction industry. J. Constr. Res. 2005, 6 (1), 71–90. (14) Rowley, H.; Lundie, S.; Peters, G. A hybrid life cycle assessment model for comparison with conventional methodologies in Australia. Int. J. Life Cycle Assess. 2009, 14 (6), 508–516. (15) Mattila, T. J.; Pakarinen, S.; Sokka, L. Quantifying the Total Environmental Impacts of an Industrial Symbiosis - a Comparison of Process-, Hybrid and Input-Output Life Cycle Assessment. Environ. Sci. Technol. 2010, 44 (11), 4309–4314. (16) Lenzen, M.; Wachsmann, U. Wind turbines in Brazil and Germany: an example of geographical variability in life-cycle assessment. Appl. Energy 2004, 77 (2), 119–130. (17) Crawford, R. H. Life cycle energy and greenhouse emissions analysis of wind turbines and the effect of size on energy yield. Renewable Sustainable Energy Rev. 2009, 13 (9), 2653–2660. (18) Bright, R. M.; Strømman, A. H.; Hawkins, T. R. Environmental Assessment of Wood-Based Biofuel Production and Consumption Scenarios in Norway. J. Ind. Ecol. 2010, 14 (3), 422–439. (19) Wood, R.; Lenzen, M. Structural path decomposition. Energy Econ. 2009, 31 (3), 335–341. (20) Baboulet, O.; Lenzen, M. Evaluating the environmental performance of a university. J. Cleaner Prod. 2010, 18 (12), 1134–1141. (21) Suh, S.; Huppes, G. Methods for Life Cycle Inventory of a product. J. Cleaner Prod. 2005, 13 (7), 687–697. (22) Heijungs, R.; de Koning, A.; Suh, S.; Huppes, G. Toward an Information Tool for Integrated Product Policy: Requirements for Data and Computation. J. Ind. Ecol. 2006, 10 (3), 147–158.

ARTICLE

(23) Strømman, A. H.; Peters, G. P.; Hertwich, E. G. Approaches to correct for double counting in tiered hybrid life cycle inventories J. Cleaner Prod. 2009, 17 (2), 248–254. (24) Department for Business Enterprise and Regulatory Reform Energy Trends; 2007; pp 1-76. (25) Lenzen, M.; Crawford, R. The Path Exchange Method for Hybrid LCA. Environ. Sci. Technol. 2009, 43 (21), 8251–8256. (26) Peters, G. P.; Hertwich, E. G. Structural analysis of international trade: Environmental impacts of Norway. Economic Systems Res. 2006, 18 (2), 155–181. (27) Huang, Y. A.; Lenzen, M.; Weber, C. L.; Murray, J.; Matthews, H. S. The role of input-output analysis for the screening of corporate carbon footprints. Economic Systems Res. 2009, 21 (3), 217–242. (28) Jungbluth, N.; Chudacoff, M.; Dauriat, A.; Dinkel, F.; Doka, G.; Faist Emmenegger, M.; Gnansounou, E.; Kljun, N.; Schleiss, K.; Spielmann, M.; Stettler, C.; Sutter, J. Lifecycle Inventories of Bioenergy. ecoinvent report No. 17; Dubendorf, Switzerland, 2007. (29) Wiedmann, T.; Wood, R.; Lenzen, M.; Minx, J.; Guan, D.; Barrett, J. Development of an Embedded Carbon Emissions Indicator Producing a Time Series of Input-Output Tables and Embedded Carbon Dioxide Emissions for the UK by Using a MRIO Data Optimisation System. Final Report to the Department for Environment, Food and Rural Affairs by Stockholm Environment Institute at the University of York and Centre for Integrated Sustainability Analysis at the University of Sydney.; Project Ref.: EV02033; Defra: London, UK, July 2008, 2008. (30) Wiedmann, T.; Barrett, J. A Greenhouse Gas Footprint Analysis of UK Central Government, 1990-2008: Report to the UK Department for Environment, Food and Rural Affairs by the Centre for Sustainability Accounting; London, UK, 30 November 2010, 2010; pp 1-41. (31) Guinee, J. B.; Heijungs, R.; Huppes, G. Economic Allocation: Examples and Derived Decision Tree. Int. J. LCA 2004, 9 (1), 23–33. (32) Huppes, G. Macro-environmental policy - principles and design; Elsevier: Amsterdam, 1993. (33) Guinee, J. Handbook on life cycle assessment. Operational guide to the ISO; Kluwer Academic Publishing: 2002. (34) International Standard Organisation. I. ISO 14041: Environmental management - Lifecycle assessment - Goal and scope definition and Inventory analysis; Geneva, 1998. (35) Reijnders, L. The life cycle emission of greenhouse gases associated with plant oils used as biofuel. Renewable Energy 2011, 36 (2), 879–880. (36) Miller, S. A. Minimizing Land Use and Nitrogen Intensity of Bioenergy. Environ. Sci. Technol. 2010, 44 (10), 3932–3939. (37) Intergovernmental Panel on Climate Change, I., Climate change 2007: the physical science basis. Contribution of working group I to the fourth assessment report of the Intergovernmental Panel on Climate Change. Solomon, S.; Qin, D.; Manning, M.; Chen, Z.; Marquis, M.; Averyt, K. B.; Tignor, M.; Miller, H. L. Eds.; Cambridge University Press: Cambridge, UK, and New York, 2007. (38) European Commission Joint Research Centre. Biofuels in the European Context: Facts and Uncertainties; 2008, 2008. (39) Searchinger, T.; Heimlich, R.; Houghton, R. A.; Dong, F.; Elobeid, A.; Fabiosa, J.; Tokgoz, S.; Hayes, D.; Yu, T.-H. Use of U.S. Croplands for Biofuels Increases Greenhouse Gases Through Emissions from Land-Use Change. Science 2008, 319 (5867), 1238–1240. (40) Fargione, J.; Hill, J.; Tilman, D.; Polasky, S.; Hawthorne, P. Land Clearing and the Biofuel Carbon Debt. Science 2008, 319 (5867), 1235–1238. (41) Intergovernmental Panel on Climate Change, I. N2O emissions from managed soils, and CO2 emissions from lime and urea application; 2006; pp 1--54. (42) Crutzen, P. J.; Mosier, A. R.; Smith, K. A.; Winiwarter, W. N2O release from agro-biofuel production negates global warming reduction by replacing fossil fuels. Atmos. Chem. Phys. Discuss. 2007, 7 (4), 11191– 11205. (43) Smeets, E. M. W.; Bouwman, L. F.; Stehfest, E.; Van Vuuren, D. P.; Posthuma, A. Contribution of N2O to the greenhouse gas balance of first-generation biofuels. Global Change Biol. 2009, 15 (3), 780–780. 2477

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478

Environmental Science & Technology

ARTICLE

(44) JEC - Joint Research Centre-EUCAR-CONCAWE collaboration, JEC Biofuels Programme -WTW results VERSION 3. In 2008; p ROFX_HY. (45) Ecoinvent. Lifecycle Inventories of Bioenergy; 2007. (46) Nemecek, T.; Heil, A.; Huguenin, O.; Meier, S.; Erzinger, S.; Blaser, S.; Dux, D.; Zimmermmann, A. Lifecyle Inventories of Agriculture Production Systems. In Swiss Centre for Lifecycle Inventories: 2007; Vol. Ecoinvent Report Number 15. (47) Intergovernmental Panel on Climate Change. 2006 IPCC Guidelines for National Greenhouse Gas Inventories: 2006; Vol 4, Chapter 5. (48) Institute of Applied Ecology Bioenergy GHG Emission Balances including Direct and Indirect Land Use Change Effects. http:// www.bioenergytrade.org/downloads/bru07indirectlucfritzsche.pdf (accessed month day, year). (49) Kim, H.; Kin, S.; Dale, B. Biofuels, Land Use Change, and Greenhouse Gas Emissions: Some Unexplored Variables. Environ. Sci. Technol. 2009, 43, 961–967. (50) International Energy Agency. Sustainable Production of Second Generation Biofuels; Potential and perspectives in major economies and developing countries; 2010. (51) European Commission. Biofuel-the Way Forward; 2008. (52) OECD Observer. Biofuel: A Second Chance. 2010.

2478

dx.doi.org/10.1021/es103410q |Environ. Sci. Technol. 2011, 45, 2471–2478